Cell loss rate estimation based on neural network for call admission control in ATM networks

This paper proposes a neural network based cell loss rate estimation method for the real time call admission control (CAC) in ATM networks. Cell loss rates data calculated by the non-parametric method were adapted to optimize the three layer perceptron. By adjusting the connection strength between neurons in the model, cell loss rates can be effectively derived from average cell rates and peak cell rates in the ATM networks. Evaluation results suggest that the proposed method is useful for high-speed ATM CAC in multimedia traffic environments.